When for instance a CO.sub.2 analyzer is used in measurement of respiration air, a problem is imposed by the water vapour present in the exhaled air. Since the temperature in the sample duct is lower than human body temperature, the water vapour will condense in the measuring instrument and the measurement may fail due to introduction of water droplets in the measuring chamber.
To the end of removing water from the sampled gas there has in gas analyzers of prior art been used a water separator wherein the condensing water is collected in a small cup, which is emptied as needed. This design solution of prior art is encumbered by the following drawbacks. The ascent time in the measurement increases, which is not desirable since the analyzer should be able to follow the changes of CO.sub.2 content during the different phases of respiration. For this reason endeavors should be to minimize the volume of the cup, but this causes the need for frequent emptying. If one forgets to empty the cup in time, the result will be failure of the measurement.